DE3431947A1 - ARRANGEMENT FOR GENERATING A CONTINUOUSLY SCANED TELEVISION IMAGE FROM SIGNALS OF AN INTERLOCK SCAN - Google Patents

Description

RCA 79 386 Ks / Ri

U.S. Serial No. 527.769

Filed: August 30, 1983

ROA Corporation New York, N.T., V.St.v.A.

Arrangement for generating a continuously scanned television picture from signals of a Interlace scanning

The invention relates to measures to reduce the Visibility of the so-called "line crawl" in a reproduced with continuous scanning TV picture coming from one for field-by-field scanning interlaced signal is generated.

Recently there has been an interest in high-resolution Television systems in which the reproduced picture has a greater horizontal and / or vertical resolution the appearance on large screens and projection screens to improve. Since such systems are unfortunately not compatible with current color television standards, one has considered the image reproduction in conventional normal definition television such as NTSO or PAL television to beautify without having to fundamentally change the usual television standards. One of the ways that have been proposed to create such an "upgraded" NTSC or PAL system is to use im

- 10 -

Television receivers interlaced scanning of partial frames Replace with continuous image scanning asu. In the NTSC system, for example, an incoming TV signal initially 262/2 scanning lines of the picture grid which appear in a field interval of 1/60 seconds, followed by a second group of 262V2 lines, which are in the grid with the lines of the first Group are nested (interlaced) and also appear in a field interval of 1/60 seconds, so that in a total time of 1/30 seconds a "monochrome" frame of 525 lines is formed. The term "monochrome" is intended to indicate the fact that the 1/30 second frame is not a complete one Repeat cycle of the phase position of the color subcarrier relative includes the horizontal sync pulse. The period within which the phase position of the color subcarrier changes completely repeated, is called a "color" frame and comprises two monochrome frame intervals (1/15 seconds) in the case of the 525-line NTSC system that works with a field frequency of 60Hz. The one working at 50Hz 625-line PAL system comprises a full color image, four monochrome full image intervals (1 / 6.25 seconds).

A progressively scanned image from a signal to form an interlaced raster requires additional scan lines during each field are provided. For this purpose, the signals, which each represent a partial image, can be increased by the duration of a Delay the field period and then in the correct order with the lines of the next field reproduce. The lines of the previous one will have an effect Sub-picture inserted between the lines of the current sub-picture. This method has the advantage that the flickering and other disturbance phenomena occurring during image movement as well as the line creeping are reduced < The disadvantage, however, is that a field memory is required to store all lines of a field for the duration of one

- 11 -

to save ηes sub-image interval. Such field memories are expensive and use considerable power.

One way to increase the number of lines in a television picture is to simply repeat each horizontal line, as described in U.S. Patent Application No. 359,612 which is in the name of R.A. Dischert was submitted. This method comes with a line memory and with simple electronics.

The use of line memories for Delaying each line of the incoming signal by a time sufficient to perform an interpolation, in order to generate signals which represent interpolated raster lines between the lines of the current field. The easiest way to do this is by averaging the signals of two adjacent lines of a field in order to get through linear interpolation to obtain a signal that represents an estimate of the signal that is the spatial represents the intervening line of a temporally adjacent partial image offset by interlacing. These estimated signals are simply inserted between the unchanged lines of the current field. It is also known to perform quadratic interpolation using memory devices save the mehifc as one line. All of this is in the U.S. Patent Application No. 300,227, filed Sept. 8. 1981 in the name of K.H. Powers was submitted. If one uses an interpolation in order to estimate intermediate lines and thereby an image reproduction of To enable interlaced video signals in double faster continuous scanning is called "pseudo-continuous" Scanning. The pseudo-continuous scan leads to the disappearance of line creeping and reduces movement-related disturbances as well as that Flicker. It can also result in a loss of vertical spatial image fineness.

- 12 -

Fig. La shows a perspective view of an im Interlaced scanned grid for the NTSC system with 525 ab t ζ t ζ t s, of which for the sake of clarity only some are shown. The scanning of the grid begins at point 1 in the upper left corner of the grid with line 1, runs to point 11 during a line interval, then returns to one Return to point 2 to start a second line ending at point 12. The scanning continues in this way, until successively 262 lines of the first field are scanned. As shown, line 262 ends at the point 14 on the right side of the grid. The first drawing file ends with the scanning of the first half of the line 263-Die Scanning of the second field begins with scanning of the last half of line 263, which ends at point 15. After the return to the left, lines 264 to 525 are scanned, those between the lines of the first field lie nested. Scan line 525 ends on Point 16. The scanning then begins again with line 1 in order to periodically move in the manner described to repeat.

FIG. 1b shows the periodic sampling according to FIG. 1a in a torn apart form to also represent the dimension of time. As can be seen from this figure, the first field, field no. 1, is scanned with 262V2 lines during the time interval TO. During the time interval T1, which is 1/60 seconds later than the interval TO (in the NTSC system), the last half of line 263, ending at point 15, is scanned. The second, interlaced Field # 2 is obtained by scanning lines 264 through 525 completed and ends at point 16. The scan continues during the time interval T2, which is 1/30 seconds later than the interval TO and in which the sub-image no. 3 is scanned in the same way as the sub-image no. 1. This sequence is constantly repeated.

- 13 -

dig further, i.e., sets the pattern shown in Fig. 1b continues infinitely to the right.

FIG. 2 is a spatial / temporal representation of the Grid lines according to FIG. 1. FIG. 2 corresponds to a view of the pattern according to FIG. 1b in the direction of view the X-axis. In Fig. 2, the scanning lines can be seen in plan view of their end and represented by dots. The scan lines of the fields of odd ordinal number ("odd" fields) are filled with black Dots and the scanning lines of "even" partial images are represented by empty, white dots in the middle, as in FIG Fig. 1. The time between successive fields is given as 1/60 seconds, but it can also make 1/50 seconds or any other duration. The distance measured in the vertical direction (Y direction) between a scan line and the adjacent scan line of the following field is shown as dimension S, and the vertical distance between the location of a scan line of a field and the adjacent scan lines of the same sub-image is equal to 2S. The adjacent scan line of the next field is in the center between the scan lines of the current field.

FIG. 3 a shows one designated as a whole by 300 Fourier transform of those going in the vertical direction spatial / temporal representation according to FIG. 2. The abscissa is measured in reciprocal values of time, that is to say in units a "Zeitfreauenz" ft, and the ordinate in reciprocal values the distance or length, i.e. in units of a "spatial frequency" fy. The spatial frequency can be measured in Periods per picture height, which is when a screen of a certain size is viewed from a certain distance can also be expressed in periods per degree of the viewing angle, as Adelson et al. in an article entitled "Modeling the Human Visual System ", which is described in Volume 27, No. 6 (November / December 1982) of the journal RCA Engineer

The desired signal in the vertical direction at any given time (i.e. with the time as a constant value) is scanned by the raster lines at intervals 2S according to FIG. Therefore According to the Nyquist criteria, the signal component belonging to the time frequency ft = 0Hz in FIG. 3a can extend in the vertical direction only up to fy = -i / 2S · Those Parts of the signal that lie at spatial frequencies that are concentrated by multiples of ^ 1 / 2S are repeated Spectral information resulting from scanning performed with 2S. These terms represent visible, undesirable "phenomena in the picture. At OHz, for example (i.e. for a constant image) the structure of the raster lines presents itself as an "appearance" at ^ 1 / S.

When you create a non-scanned white sub-image that periodically lights up every 1/60 of a second and goes dark then one gets a phenomenon which is expressed in the transformation plane according to FIG. 3a by points at ft = i60Hz along the frequency axis represents fy = O. This phenomenon is known as large area flicker.

There is also another discreet appearance that is both Spatial frequency components as well as time frequency components has and in the Fourier transform plane according to Fig.3a at points with the coordinates ft = i30Hz, fy = ± 1 / 2S. This phenomenon is known as line crawl and is caused by the nesting of the lines successive partial images. To explain the physical form of this phenomenon, consider the case that an eye at a location as shown at 210 in Figure 2 is scanning the image in a vertical direction. At certain speeds of this eye scan, the successive raster lines appear to be more vertical Move direction. This line creep is the psychovisual perception that the lines are spatially move as a function of time, and results from that

- 15 -

the eye follows a spatial / temporal path like him represented by the dashed line 212 in FIG. 2 is.

The pattern shown in lig * 3a represents transforms of Components represent how they appear when using a white image (represented by white rectangle 340 in Figure 3b) a field frequency of 60Hz interlaced is scanned. The spectrum of an image which has a transition between black and white is shown in FIG. 3c and designated as a whole by 350. The one mentioned Transition (shown in Fig. 3d) gives rise to sidebands or spectral components which move in the fy direction as shown by dashed lines 356 in Figure 3c. These vertical components represented the so-called line or edge flicker. If the edge or the transition between the black area 352 and the white area 354 - of the in Fig. 3d then the lines 356 spread to those with general phenomena of motion Fill in the rectangles of the spectrum 350, in which the components of edge flicker. For a better understanding, the components of Fig. 3c are in a perspective View shown in Fig. 3e, the amplitude axis standing at right angles to the axes fy and ft is drawn.

Figures 4a-h are representations for explaining Terms used in describing interpolation filters. In these figures, the The abscissa represents the vertical direction, measured in grid lines. In Fig. 4a is an arbitrary image signal 410 that ranges from 1 to 3 raster lines has a value of 1.0 units and has a value of 2.0 units in the range of 6 to 8 raster lines, with a intermediate smooth transition between these two values. The signal only exists at the points of

- 16 -

Grid lines as shown by the black dots. The signal can e.g. correspond to an image, which is black in the upper part (grid lines 1 to 3) (low signal level) and in the lower part (raster lines 6 to 8) is white (high signal level), with a transition in the area of raster lines 3 to 6. Figures 4b to 4g show some of the successive positions, which takes the response of an interpolation filter working with three taps (i.e. linearly interpolating), if this filter samples the signal over time, which in the present case is equivalent to a spatial one Is scanning in the vertical direction. The illustrated "response" of the filter (response to impulse) has three maxima or "peaks" 412, 4-14 · and 416, their mutual Distance corresponds to the spatial distance S, that is to say half the distance between successive scanning lines. Peak 414 has a "multiplier" or "value" of 1/2 or 0.5 while peaks 412 and 414 each have a value of 1.4 or 0.25. The multipliers of the various peaks in the response of the filter are chosen or normalized in such a way that they form a sum value equal to one. In this way it is achieved that the intensity of the image before or after the interpolation of lines is the same, otherwise the doubling would be the number of lines lead to double the brightness. When the filter receives the image signal 410, this has the effect of that the filter response performs a spatial scan. At the point in time, which is shown in Fig. 4b, falls The 0.5 peak 414 of the filter response coincides with raster line 1, while peaks 412 and 416 do not coincide with a raster line. The value that the signal generated by the filter will have at any point in time in the course of the scan results from multiplication of each of the signal values in the relevant Time of the peaks of the filter response can be detected, in each case with the multiplier of the relevant Peak, and subsequent summation of the

- 17 -

values. In the filter position shown in Fig. 4b, the peak 414 detects a signal with the value 1.0, and tips 412 and 416 each sense a null signal. The value of the filter output signal at this point sition is:

(0.25. O) + (0.5. 1.0) + (0.25. O) = 0.5.

This value supplied by the filter is with point 420 shown in Fig. 4h. The filter continues scanning, half a line spacing later that of FIG. 4c to assume the position shown. In this position, peak 414 does not detect a signal during the filter response the peaks 412 and 416 each detect a signal with the value 1.0 at the locations of the raster lines 1 and 2.

The value of the filter output resulting in this case is calculated as follows.

(0.25. 1.0) + (0.5 - 0) + (0.25. 1) = 0.5,

which is shown at 422 in Figure 4h. In the further course of the scan, the filter arrives one after the other various vertical positions, some of which are shown in Figures 4d-g. In the course of each measurement at a distance 2S, the filter supplies an output signal twice. One output signal appears when the middle peak 414 of the filter response detects the signal value at one raster line, and the other signal appears, when peak 414 is midway between two raster lines so that peaks 412 and 416 are adjacent Capture raster lines. If the tip 414 detects a signal, then the output of the filter is proportional to the actually detected signal. If tips 412 and 416 each sense a signal, then is the filter output signal proportional to the mean value of this two detected signals. In Fig. 4, the value of the product of the multiplier is each shown

- 18 -

Peak with the respective detected value of the image signal 4-10 entered below the respective peaks. The filter output represents the actual raster lines and there are additional raster lines in between, which are interpolated by averaging. Other filter characters can be recorded represented in the same way by a series of spaced response peaks, where the output of the filter is the sum of the various instantaneous products obtained by multiplying the result in individual peak values with the respective detected signal value.

The present invention relates to an arrangement for generating a continuously scanned television picture of signals representing an image scanned in a raster in which lines of first (even) fields have a vertical distance of 2S and are interleaved with the lines of second (odd) fields According to the invention, the lines of the signal to be applied to an image display device are determined by means of a filter device which has such a characteristic in the vertical direction that for each line of the incoming signal a plurality of lines is generated at the same time. The reproducing device gives the plural of the filtered lines in adjacent raster lines again.

The essential features of an arrangement according to the invention are described in claim 1. Advantageous configurations of the invention are characterized in the subclaims.

The invention is explained in more detail below using exemplary embodiments with reference to drawings.

Figures 1a-b (already discussed) show in perspective View of the scanning of a television grid;

- 19 -

Fig. 2 (already dealt with) is a spatial / temporal one Interlaced raster scan diagram;

Figures 3a-d (already dealt with) show in the Fourier transform planar phenomena which, in the raster scan according to FIG. 2, arise at different Result in image content;

Figures 4-a-h (already discussed) show the value such as e.g. the amplitude of a video signal as a function of time or the vertical position in the raster with an interpolation filter in different time positions as well as the resulting interpolated Waveform;

Figures 5 and 7-10 show different filter responses and the associated filter curves;

FIGS. 6a-6p show various functions and filters in the image plane and their transforms in the Spatial frequency plane;

Fig. 11 is a spatial frequency diagram for explaining the coverage of the adjacent band; '

Figure 12 is a block diagram of a television receiver producing a continuously scanned image in accordance with the invention;
30th

13 shows a block diagram and the frequency diagram of an interpolating and time-pressing device, which can be used in the arrangement of Figure 12;

14 shows a block diagram of a television receiver incorporating a switching interpolation filter in accordance with an embodiment of the invention; _{? 0}

Fig. 15 shows scanning waveforms of the image display device in the arrangement of Pig. 14;

FIGS. 16 and 17 show interpolation filters which are used as a substitute in the arrangement according to FIG can;

Fig. 18 shows the spatial frequency response of an ideal filter

and a filter according to the invention; 10

Fig. 19 shows part of a distribution showing a filter frequency response;

Fig. 20 shows the shape of a signal and the resultant Signal shape after the action of the frequency response according to FIG. 19;

Fig. 21 shows a signal following a step function and a desired form of the signal response a filter;

22 shows a distribution truncated in the direction of the spatial axis and its relation to a distribution according to the invention Filter for obtaining a response signal as shown in FIG.

The curve 510 shown in Fig. 5a represents the so-called "Spatial frequency response", i.e. the amplitude transmission ratio as a function of spatial frequency, for one with three taps working filter which the. already shown in Fig. 4 and in Fig. 5b repeated response has function. Under answer function or short "answer" of a filter is in the present case the amplitude of the impulse response as Understand the function of vertical distance. The spatial frequency response curve 510 is thus a Fourier transform the filter response shown in Fig. 5b. This is easy to understand when one thinks that the Fourier

- 21 -

Transform of a pair of pulses spaced by the distance 2S is a cosine curve and that the addition of an additional third pulse in the middle between the pulses of the pair offsets them Cosine curve around the amplitude of the third pulse means, as it is on page 33 of the work "The Fast Fourier Transform "by E. Oran Brigham (Prentice Hall, 197 * 0) is. The curve 510 is thus half an amplitude reduced cosine function with an amplitude offset of 0.5. The amplitude of the frequency response curve 510 is equal to O at a spatial frequency (fy) of 1 / 2S.

In FIG. 5c, the Fourier transform 300 according to FIG. 3 is shown again and next to it the spatial frequency response 510. The maxima (amplitude equal to 1.0) of the frequency response 510 shown are at fy = O, 1 / S and -1 / S. Further maxima, which are no longer shown in FIG. 5c, appear at ii / S, ± 2 / S, ^ 3 / S, etc. The zeros of the frequency response (amplitude equal to zero) appear at those vertical spatial frequencies that are shown in lie in the middle between the maxima, in particular at those frequencies around which the phenomena of line creep are centered, namely at -1 / 2S. The course of the frequency response 510 with its maximum at fy = O leads to the fact that the lower spatial frequencies are maximally transmitted at all time frequencies, while the transmission at fy - 1/4-S takes place with half the amplitude and the transmission factor at spatial frequencies 1 / 2S, where the components of line creep are centered is zero.

Desired signals that occur at spatial frequencies close to 1/4-S, are thus attenuated by about 6dB, and therefore the filter in question causes an undesirable reduction of vertical details.

The curve 510 of the spatial frequency response is shown in FIG. 5a only over the range from fy = 0 to fy = 1 / S. Naturally the spatial frequency response extends in reality over

- 22 -

the area ico in repeating periods of which only one is shown. Therefore the maxima appear in reality at ± 2N / 2S, where N = 0,1,2 ..., and between these maxima are the zeros. If in the following different filter curves in the spatial frequency level are dealt with, the presentation and explanation is limited to a few or a few periods, but applies correspondingly for the whole spectrum.

Figures 6a-p illustrate the principle according to which the Frequency components of the line creep are eliminated as far as possible by means of a suitable interpolation filter should. Fig. 6a shows with curve 410 a typical image signal, such as that coming from a camera, for example by the amplitude as a function of the vertical distance. Figure 6b shows the Fourier transform 612 of the signal 410 by representing the amplitude CO (fy) as a function of the spatial frequency. This spectral function 612 includes, because of the soft transition in signal 410, few high spatial frequencies.

6c shows the amplitude of the signal 410 as a function of the spatial distance when scanning through the raster lines of a first television field. This scanned Version consists of a number of pulses that appear at periodic intervals of length 2S and the amplitudes of which are derived from the amplitude of the signal 410. 6d shows the spatial frequency spectrum 616 of the signal shown in Fig. 6c. The spectral distribution 616 has maxima in spatial frequency intervals 1 / 2S, which is the consequence of sampling at spatial intervals 2S. The spectrum 616 has a part 618, the corresponds to the spectrum 612 of the signal 410, and further partial spectra 620, 622, the counterparts of the partial spectrum 618 and appear at ii / 2S, ii / S ... The counterparts 622 do not overlap the sub-spectrum 618, while the counterparts 620 with both the sub-spectrum 618 and also overlap the counterparts 622.

- 23 -

Fig. 6β shows the response function of one with three taps working interpolation filter, as it is already shown in Figures 4 and 5 and one Scan in the direction of the arrow to capture portions of the signal 615. Fig. 6f shows the Fourier transform this response function, corresponding to the upwardly offset cosine curve shown in FIG. FIG. 6g shows the output signal 630 which is provided by the filter when the response function 4-12-416 receives the signal 615, as described above in connection with FIG. 4. Fig. 6h shows the transform. mated or spectral function 632 of the signal 630. This Spectral function 632 has maxima at O, ii / S, which is what one looks at can explain that the signal 630 is similar to a version of the sampled at spatial intervals S Image signal 4-10. Alternatively, the transform 632 can also be considered as the result of multiplying the spectrum 616 by the transfer function 510 of the Filters * It should be mentioned that the amplitude maxima of the transfer function 510 correspond to the peaks of the partial spectra 618 and 622, so that the peak values in the spectrum 632 stay the same. On the other hand, the zeros at ii / 2S correspond to the transfer function 510 of the filter positionally corresponds to the peaks of the partial spectra 620, so that these peaks are practically reduced to zero will. In the areas on both sides and close to the spatial frequencies ii / 2S has the transfer function 510 of the filter Values that are small but not equal to zero, so that a small proportion of the information falling into the partial spectra 620 still exists in spectrum 632. This unwanted remainder of the partial spectra 620 in principle occupies the higher-frequency parts of the partial spectrum 618, i.e. those parts which are closer to the spatial frequencies fy = ii / 2S than lie at the zero frequency fy = O.

Figure 6i shows the signal 634- which occurs when sampled of the image signal 4-10 results in a second field which

ches is interlaced with the first partial image. The sampling points 636 appear at intervals 2S, and their amplitudes depend on the magnitude of the image signal 410, as in the case of FIG. 6c. The sampling points according to FIG. 6i are, however, offset by the vertical distance S compared to the sampling points according to FIG. 6c. 6j shows the Fourier transform or spectral function 638 of the signal 634. Because of the aforementioned offset of the sampling points, the parts 620 in the spectral function 638 have opposite amplitudes compared to parts 618 and 622, which are centered around the spatial frequencies 0 and ^ 1 / 2S are. Figures 6k and 61 are repetitions of Figures 6e and 6f to make the drawing easier to read. 6m shows the output signal 640 of the interpolation filter when its response function 412-416 samples the signal 634. The spectrum 642 in Fig. 6n is the Fourier transform of the signal 640. This spectrum 642 can be viewed as the result of a multiplication of the spectrum 638 of Fig. By the transfer function 510 of Fig. 61. As in the previous case, the sub-spectra 618 and 622 at the frequencies 0 and ^ 1 / S, etc., essentially unaffected, while the parts 620 at the frequencies ^ 1 / 2S, i3 / S (no longer shown) are attenuated. The signal 646 and the associated spectrum 650 in FIGS. 6o and 6p represent the sum of the signals 630 and 640 or the sum of the spectra 632 and 642. As can be seen, the amplitude of the signal 646 is increased. The amplitudes of the parts 618 and 622 of the spectrum are also increased (which cannot be seen in FIG. 6p, because in this figure the scale factor in the CO (fy) direction has been changed from CO to 2 cd for reasons of space) the positive and negative residues of the damped parts 620 cancel each other out. This cancellation also leads to the elimination of line creep phenomena, because the transfer function of the filter is zero for the spatial frequency coordinates fy = £ i / 2S.

- 25 -

By eliminating the phenomena responsible for line creep in the signal, the line creep effect becomes eliminated in large, evenly illuminated areas of the image. However, the line flicker will not completely eliminated, nor the more general motion effects · To the same extent as the line flicker is reduced, vertical details in the image are also lost.

As mentioned, one operates with three taps Filter that has the response function shown in Figures 4 and 5 512, an interpolation by taking the mean value of neighboring lines. Fig. 7a shows the Response function 710 of a filter that works with two taps, both of which are weighted with a factor of 1/2 and have a distance S from each other. The use of such a filter is equivalent to a simple one Repeat every raster line. Fig. 7b shows the Fourier transform 712 of response function 710, that is the spatial frequency response of the filter. This frequency response is a simple cosine curve with no amplitude offset. Because of the lack of an amplitude offset, the Cosine curve 712 at fy = 1 / 2S a phase reversal from positive to negative phase. The frequency response has with this Spatial frequency 1 / 2S has the value zero, so that the line creep as in the case of the filter operating with three taps 5 is eliminated. Furthermore, the filter according to FIG. 7 has a very good transmission behavior for low Frequencies. The loss of high-frequency signal components in the area between fy = i / 4S and 1 / 2S is somewhat smaller than in the case of the filter working with three taps, but the damping of motion phenomena also lower (the transmission factor near fy = 1 / 2S is higher). Hence, the two-tappet leads "line repeating" filters for an image that has a slightly better sharpness, but also a little more movement effects shows as an image created using the three-way

- 26 -

1431947

taps working "averaging" filter generated will. It should be noted that the frequency response 712 has a positive part from fy = 0 to 1 / 2S and a negative part between 1 / 2S and 1 / S. The negative part of the frequency response is undesirable because it means the effect is a vertical shift of the affected parts of the repeat signals. That is, in a picture that is below 7 is generated, the positions of the scanning-related phenomena are higher Spatial frequency offset.

Fig. 8a shows the response function 810 of another filter which operates with four taps, their mutual Distance corresponds to the vertical distance dimension S.

The weight values of the taps, expressed using of a parameter p, are p, 1/2-p, 1/2-p, p. The parameter ρ can be chosen equal to zero, and in this In this case, the Fourier transform of the response function results in a spatial frequency response as shown by curve 812 is shown in Fig. 8b and that of the frequency response 712 of the filter operating with two taps according to the figures 7a and 7b correspond. A parameter value ρ of -1/8 results in the frequency response 814, and with p = -1 / 4 one obtains the Frequency response 816. It can be seen that increasingly negative values of ρ lead to an ever greater increase in the amplitude of the Frequency response at spatial frequencies above zero and below 1 / 2S, i.e. in the image area of the spatial frequency spectrum. This improves the apparent vertical resolution of the image. The line creep effects at fy = i / 2S are dampened, because the frequency responses 812 to 816 all have the value zero at this spatial frequency. With such a filter the frequency response undesirably has a negative sign (negative phase) in the range between 1 / 2S and 1 / S. This results in the above-mentioned shifts within the relevant range of spatial frequencies.

9a shows the response function 910 of a filter,

- 27 -

that works with five taps that are spaced apart according to the distance S and are weighted with values p, 1/4, 1 / 2-2p, 1Λ »P. The curves 912, 914 and 916 in Figure 9b show the Fourier transforms the response function of the filter for p = 0, -1/8 and -1/4. The curve 912 for p = 0 corresponds to the curve 510 of the filter operating with three taps according to FIGS. 5a and 5b. All frequency response curves shown in Fig. 9b go through zero at 1 / 2S so that they eliminate the line creep effect. For increasingly negative values of ρ the amplitude of the frequency response in the range between fy = O and fy = 1 / 2S is increased more and more, so that the The transmission dimension in this spatial frequency range is greater than with fy. = O. There is no phase reversal in the area between 1 / 2S and 1 / S. This filter has a better effect than the filter operating with two taps according to FIG. 7, because the attenuation is greater near fy = £ i / 2S, which it can be deduced from this that the slope of the frequency response at 1 / 2S is equal to zero. The signal is in the field between 0 and 1 / 2S, and there is no phase reversal in the range from 1 / 2S to 1 / S.

Fig. 10a shows the response function of a square interpolation filter with seven taps, which is generally corresponds to the filter of U.S. Patent Application No. 300,227 mentioned above. The seven taps are weighted with p, 0, 1/4-p, 1/2, 1/4-p, 0, ρ and have a mutual distance corresponding to the distance measure S. In Fig. 10b, the transforms of Filters for p = 0, -1/16 and -1/4 are shown. For p = 0 the filter is reduced to working with three taps end, averaging filters according to FIGS. 5a and 5h. However, the filter works for negative values ρ a signal increase with a positive phase in the range from fy = 0 to about fy = ± 1 / 4S and a signal increase with a negative phase Phase in the range from fy = ii / 4S to ii / 2Sist. thats why this filter is not optimal for the purpose of image sharpness

to "preserve" while at the same time dampening the effects of movement.

It has already been mentioned that the process of filtering out ^ß aumfrequenzen close to II / 2S inevitably leads to the attenuation of desired signal components, because of the characteristics of spatial frequency filters. While this is undesirable, it does not result in as much image degradation as one might think. The reason for this is a psychovisual phenomenon which can be described as "covering the spatial frequencies of the adjacent band". This phenomenon leads to spatial frequency information being obscured by other spatial frequency information occupying the same area of the image if the spatial frequencies of the two items of information are within half a spatial frequency octave (spatial frequency ratio 2: 1) apart. 11a illustrates this principle. The spatial frequency spectrum of the desired signal is represented by an envelope 1110 which extends from fy = 0 to almost fy = 1 / 2S. The repetition or noise spectrum centered around 1 / S, which results from the raster scanning with a line spacing S, is the hatched area under the envelope curve 1112. Those signal components which are close to 1 / 4S are covered by noise components close to 1 / 2S, and the signal components in the range from 1 / 4S to 1 / 2S are also covered by the noise components near 1 / 2S, with the covering being greater for the signal components located near 1 / 2S because these components are covered by the noise components located further away at 1 / S be covered. Similarly, the noise components in the range from 1 / S to 3AS are masked by the signal components. The components in the spatial frequency range from 1/4-S to 3/4-S are therefore reproduced with reduced visibility. Fig. 11b applies to an image that. contains no signal in the range from fy = 1 / 3S to 2 / 3S and therefore not only does not appear worse than the image on which FIG. 11a is based, but its limited spectrum is actually better because it

- 29 -

There are sequences in the range from 1 / 4S to 1 / 3S that are not covered. This result is important because it indicates that images of a continuously scanned scene differ little in their appearance when the interpolating Filters attenuate the signal in the frequency range above 1 / 3S. The filters of Fig. 10, however, attenuate the spatial frequency spectrum already at 1 / 4S around 6dB, so over a larger range than it is with regard to the "occlusion of the adjacent band "would be required.

Fig. 12 shows an interlaced television signal receiver which is a continuously scanned Image generated using an interpolation filter having a response function as shown in Fig. 9a. In the Arrangement according to Fig. 12, KTSC signals transmitted via radio, which are modulated onto a carrier by an antenna 1210 and collected on a tuner and IF amplifier given, which are shown collectively as block 1212 and in which the signals are converted to an intermediate frequency (IF) will. The IF signals are sent to a control circuit 1214 for automatic gain control to influence the gain of the tuner and the IF amplifier.

The IF signals are also fed to a detector 1216 which demodulates a composite color television signal generated in the baseband. The 4.5 MHz intercarrier containing the audio information of the signal is placed on an audio channel, the one intercarrier amplifier 1218, an FM de-

30. modulator 1220, a sound amplifier 1222 and a loudspeaker 1224 contains. The baseband television signal from detector 1216 is also a sync pulse separator 1226 applied, which separates the vertical and horizontal sync signals V and H and also a burst identification signal BF which is applied to a burst gate circuit 1228. The horizontal sync signals H are sent to a circuit 1230 for automatic frequency and phase control (AFPR) assigned

leads to stabilized horizontal sync signals H. generate which on a scaling circuit 1232 for frequency division be coupled to vertical frequency, as is known per se. The stabilized horizontal sync signals H are also fed to a circuit 1234, a signal of twice the horizontal frequency (2H) with automatic frequency and phase control as a timing signal generated for continuous sampling at twice the frequency. The signals from the reduction circuit 1232 are fed to a vertical deflection circuit 1236 which is a vertical deflection winding attached to a picture tube 1240 1238 drives. The signals of twice the horizontal frequency supplied by the AFPR circuit 1234 are fed to a horizontal deflection circuit 1242 which has a horizontal deflection winding 1244 on the picture tube 1240 controls.

The composite video signal from detector 1216 is fed to a luminance / chrominance separation circuit 1246, wherein the luminance component X is separated from the chrominance component C. The chrominance component is given to the burst gate circuit 1228, which couples the burst of the chrominance signal to a circuit 1248 for regeneration of the subcarrier. The regenerated A subcarrier is applied to a chrominance demodulator 1250, the components I and Q of the chrominance signal in the baseband wins.

The luminance signals Y from the isolation circuit 1246 become to an interpolation filter / time press circuit 1252, which has an answer function as it does with 910 is shown in Figure 9a. Circuit 1252 includes a cascade connection of four delay lines 1254-1260, each of which has a delay H of about 63 »5yus causes what corresponds to one horizontal line period in the NTSC system. Taps 1262-1268 on the delay cascade 1254-1260 couple samples of the delayed luminance

- 31 -

nals to assigned 12 dB attenuators, which are called blocks 1269-1272 are shown.

12dB corresponds to an amplitude ratio of 1: 4-, and therefore the samples of the luminance signal appear at the outputs of the attenuators 1269-1272 each with a Amplitude, which is reduced to 1/4 compared to the input amplitude is. this ratio corresponds to the value ρ = 1/4-in Fig. 9 · The output signals of the attenuators 127Ο and 1271 are applied to an adder 1273 to produce a to form an interpolated signal that is applied to an input of a time presser 1274-. A second entrance to the time press 1274- receives the output signal of a summing circuit 1276, which in turn, at two inverting inputs, the output signals of the attenuators 1269 and 1272 and at a non-inverting input the output signal of delay line 1256 receives. The signal sent from delay line 1256 to the non-inverting Input of summing circuit 1276 is not attenuated because for a value of p = -1 / 4 the value or multiplier 1 / 2-2P of the mean peak of the response function 910 equals 1 (no attenuation). Of the Time Presser 1274 · contains a multitude of delay lines that carry simultaneous parallel input signals receive and deliver time-compressed sequential output signals at twice the frequency. Such a time press is described in greater detail, for example, in U.S. Patent 4,376,957.

The signals I and Q from demodulator 1250 are also sent to an interpolator / time presser 1278 and 1280, respectively which are constructed in the same way as the circuit 1252, the time-pressed interpolated signals T, I and Q circuits 1252, 1278 and 1280 become a matrix circuit 1282 applied, which derives color control signals E, G and B therefrom, which a driver circuit 1284 control the picture tube 124-0 are fed.

- 32 -

Fig. 13 shows an embodiment of the invention which can be used as an interpolator / time presser instead of the circuit 1252 in Fig. 12 can be used. The arrangement of Fig. 13a has a response function as shown at 1301 in Fig. 13b and which corresponds to the filter according to FIG. 8 for p = -1 / 8. In Fig. 13, the non-pressed video signal becomes is applied to a cascade of H delay lines 1310-1314, each having a delay time of approximately 62.5 / us has. Attenuators 1316 and 1318 with attenuation dimensions

of 18.06 dB (corresponds to a value p = 1/8) are with the input of delay line 1310 and coupled to the output of delay line 1314. More attenuators 1320 and 1323 with an attenuation of 4.08dB each (corresponds to the value p = (1/2-p) = 5/8) are with the connection point of delay lines 1310 and 1312 or the connection point of delay lines 1312 and 1314 coupled. The output of the attenuator 1316 is connected to the inverting input and the output of the attenuator 322 to the non-inverting input of a sum- circuit 1324, and are connected in a similar manner the outputs of attenuators 1318 and 1320 with the inverting and the non-inverting input of a Summing circuit 1326 connected. The signals that appear simultaneously at the outputs of summing circuits 1324 and 1326 appear, are given to the input of a time presser 1374, which these two simultaneous input signals picks up and a Ββςμβη ^ βΐΐββ, time-pressed output signal which is generated on a matrix circuit 1280 can be given in order to combine it with other video signals for the generation of kinescope signals.

14 shows another embodiment of the invention, in which a vertical line sweep is used to create an effectively continuously scanned image by means of in- ' to generate terpolation, in which line creep effects are reduced. 14, one is shown as block 1412 Circuit arrangement that includes a tuner and IF

- 33 -

and includes AGC circuitry coupled to antenna 1412, to receive NTSC signals sent to a carrier are modulated. One to the output of the arrangement 1412 coupled detector 1416 demodulates the NTSG signal to convert to generate a composite video signal that is sent to a sync signal separator 1426 and to a Y / C separator circuit 1446 is laid. The synchronizing signal separation stage 1426 provides a burst identification signal BF which, together with the Chrominance signal C from separation stage 1426 to a chrominance processing unit 1450 is given to generate chrominance component signals I and Q. The sync signal separation stage 1426 supplies horizontal sync signals H, which is an AFPR circuit operating at the horizontal frequency 1430, in which signals at 15734.266 Hz are generated for. Acting on a vertical frequency reduction circuit 1432 and a horizontal deflection circuit 1442. The horizontal deflection circuit 1442 controls a horizontal deflection winding located on a picture tube 1440 1444 at. The subscriber circuit 1432 provides vertical control signals to a vertical deflection circuit 1436, which a vertical ab development 1438 with a 60-Hz * - Signal controls.

The chrominance signal 0 from the Y / C separation circuit 1446 becomes to a filter 1452 which has the frequency response 1810 shown in FIG. 18 and which is a cascade connection of two H delay lines 1454 and 1456. An attenuator 1460 receives an input signal from the output of delay line 1454 and provides an output signal attenuated by a factor of (1/2-p). Another one Attenuator 1462 attenuates the signal that is also given to the input of delay line 1454 the factor p. A third attenuator 1464 attenuates this signal coming from the output of delay line 1456 also by a factor of p. The output signals of the attenuators 1462 and 1464 are each shared with the output signal of the damping energy 1460 to one sum

- 34 -

encoder circuit 1466 or 1468 given. A switch 1470 alternately selects the output signals of the summing circuits 1466 and 1468 under the control of a limiter and switch control unit 1482, as indicated by the dashed line Line 1472 is indicated. The output of switch 470 is a luminance signal T, which is common with I and Q signals generated by I and Q pilterns 1476 and 1478 come, a matrix circuit 1480 is applied.

The switch 1470 is controlled by the control unit 1482 at a frequency that is an even multiple of half that Horizontal deflection frequency is. In the embodiment 14, the switch control unit operates at 1024 times half the line frequency, which is slightly more than 8 MHz is. The desired switching frequency is set by a Maintain phase-locked loop (PLL), which controls the limiter and control unit 1482, a Binary counter 1482 acting as a frequency divider, a phase detector and filter circuit 1486 and an oscillator 1488. An output of the limiter and switch control unit 1482 is also connected to an auxiliary vertical deflection circuit 1490 connected, in turn with an auxiliary vertical deflection winding 1492 is coupled.

When the output signal of the limiter and switch control unit 1482 is in a first state, then switch 1470 is in a first position, and the auxiliary vertical Deflection circuit 1490 energizes the HiI fs vertical deflection winding 1492 to deflect the beam up the picture tube by a distance * j S. When the output of the limiter and switch control unit 1482 is in the other state, switch 1470 is in the other position, and the beam will go down over a distance

• p S deflected.

The control of the switch with a frequency that corresponds to an even multiple of half the line frequency.

- 35 -

T speaks causes full during each line interval Switching periods occur so that there is no progressive phase shift from line to line. This is desirable to that shown in Figure 15b To obtain scanning patterns. The Pig. 15t> shows a scan pattern, how it arises when the auxiliary vertical deflection winding is driven with an even multiple of half the line frequency. In Fig. 15b show the solid lines Lines the scanning of odd fields, while the scanning of even fields by the dashed Lines is shown. It can be seen that the positive deflection of each period is aligned with that of the adjacent lines, so that with each horizontal deflection effectively two lines are scanned which are at a vertical distance S from one another. To this Way, the number of lines for each sub-picture is doubled. During the positive excursions of each scan, one of which is designated 1510 in Figure 15h the switch 1470 shown in FIG. 14 is in its left position, so that that shown in the interval ^ Q-t ^ Output signal is derived from the input and output of delay line 1454 in FIG. While of the next interval, from t ^ -t ^, causes the 8 MHz square wave a negative vertical deflection of the Auxiliary deflection, and at the same time controls the switch control unit 1482, switch 1470 to its right position (not shown). The luminance output signal is in this switching state of filter 1452 is the sum of one from the input and one from the output of delay line 1456 derived signal. In this way, the desired picture reproduction sequence is obtained. Fig. 15a shows in comparison FIG. 15b shows the pattern of a conventional deflection without line wobble.

16 shows the block diagram of a filter 1610, the one response function corresponding to function 510 in FIG

- 36 -

Fig. 5b has (weight values 1/4, 1/2, 1/4 at the taps), and that instead of the filter 14-52 (or 14-76, 14-78) in Fig. 14- can be used. The signal applied to filter 1610 comes on 1H delay line 1612 and also via a 12 dB attenuator 1614- to a first input a summing circuit 1618 and finally also on a terminal of a switch 1620. The switch 1620 is via a control line 14-72 with a switching frequency operated at nominally 8MHz as described in connection with FIG. The output of the delay line 1612 is applied to a second input of the summing circuit 1618 via a 12 dB attenuator 1616. The output of the summing circuit 1618 becomes one the second terminal of the switch 1620 is supplied. The switched luminance signal from the output of the switch is on the matrix circuit 1480 (Fig. 14) coupled to it for image display in the manner previously described to process.

FIG. 17 shows a filter 1700 which has a response function corresponding to function 910 in FIG. 9 and instead of filter 1452 (1476, 1478) in Figure 14 can be used. In the filter of FIG. 17, the input signal becomes a cascade connection of 1H delay lines 1710, 1717 given and also via an attenuator 1714- with the weight factor ρ to an input of a summing circuit 1726. The one delayed by delay line 1710 Signal is via an attenuator 1716 with the weight factor (1 / 2-2p) to a second input of the summing circuit 1726 and also via a 12dB attenuator 17I8 to a non-inverting input of a summing circuit 1724. The signal from the output of the delay line 1712 has a 12dB attenuator 1720 given to a second input of the summing circuit 1724 and also via an attenuator 1722 with the weight factor ρ to an input of the summing circuit 1726. The outputs of summing circuits 1724 and 1726 are with

- 37 -

connected to the terminals of a switch 1728, which is either selects the sum of two signals delayed by 1H and attenuated by 12dB or the sum of one signal delayed by 1H and according to the weighting factor (1 / 2-2p) attenuated signal with two further signals, one of which undelayed and the other is delayed by 2H and both of which are damped according to the weighting factor ρ. The value ρ can be negative, with inverting inputs of the summing circuit 1726 would have to be used. The toggled signal from switch 1728 is applied to the matrix circuit 1480 as described above.

As mentioned above in connection with Pig. 11, the apparent quality of a scanned Image can be improved by eliminating those frame rate and image frequency components that subject to the mentioned "concealment by the adjacent band". These are spectral components, spaced from one another within an octave of spatial frequency. 18 shows the spatial frequency spectrum 1810 of a Filters that have a frequency response amplitude at the spatial frequencies 1 / 3S and 2 / 3S, which are in a ratio of 2: 1 to each other of 1/2 (corresponds to -6dB). At the spatial frequency 1 / 2S is the amplitude of the filter frequency response equal to zero, so that line creep is eliminated. Imagine that with additional taps and so that additional sections of the filter a rectangular frequency response curve is obtained, as it is shown by the broken line 1812 in Fig. 18 so that an even better picture can be displayed can. The response function of a filter with an infinitely steep slope between the pass band and stop band follows one (sin x) / x distribution as shown by curve 1710 in FIG. 19. The curve 1910 has one main lobe 1912, a negative first side lobe 1914, a positive first side lobes 1916, positive and negative second side lobes 1918, 1920, etc .. The curve 1910

- 38 -

stretches left and right "until S = ic ^>. If the answer function 1910 is linked ("folded") with a signal according to the convolution theorem, then there are prefixes and Post-oscillation in the output signal. Will the answer function 1910 e.g. with a jump function Signal convolved as it is shown with 2010 in Fig. 20, then has the filtered output signal as it shown at 2020 has a very short rise time, but pre-oscillation and post-oscillation excursions such as e.g. 2022, 2024, 2026, 2028. That means that you get a short response time, but on the other hand you also get distortions of the image signal due to large pre- and post-oscillation amplitudes.

The eye is very sensitive to such pre- and post-oscillation. If you apply these vibrations by damping or other measures to a single pre-oscillator and a limited single post-oscillation, as is shown by curve 2120 in FIG. 21b, then the image appears subjectively better than in the case of signal 2020 in FIG. 20b, although the steepness of the jump transition in signal 2120 is not so great as in signal 2020. Such reduced oscillation phenomena are obtained if one uses the (sin x) / x distribution, which the Describes the impulse response of the filter as a function of the vertical distance (i.e. the "response function" of the filter represents), shortened.

The shortened (sin χ) / x distribution (2110 in Fig. 22a) can can be used to describe the envelope of the multipliers for tapping a filter, while the distance ! between the taps is determined by the vertical distance S. This is illustrated in Fig. 22b, where the taps are arrows with an amplitude indicating the appropriate multiplier are drawn and the distance S is the abscess 3 TT ^ / 4- is equivalent. It should be noted that the in Fig. 22b drawn weight factors or multipliers of the taps do not exactly result in the sum value 1. It

- 39 -

it can be useful to set the weighting deviating from the actual values of the (sin x) / x function in order to the effects resulting from the shortening on the zero amplitude at 1 / 2S and the one value at 1 / S = Result in 0. It has been found that between a filter with five taps and a filter with seven taps there are few differences in terms of frequency response. To only one single Vorwinger and Hachwinger To get it seems necessary to the weight factors the taps monotonically diminish over the main lobe and then curve over at least part of the first To follow the lateral lobe, but no further than over the first lateral lobe. If more side lobes are involved, Then there are transient phenomena in the output signal of the filter. Which cause a transition on both sides light and dark areas become visible in the image. Interpolation filters according to the invention thus lead to subjective ones Image enhancements when considering the weight factor or multipliers of the taps according to a shortened (sin χ) / x distribution dimensioned and the cutoff frequencies the filter is selected in such a way that the area of the adjacent band from fy = 1 / 3S to 2 / 3S is attenuated. The vote the exact hatred of damping at fy = 1 / 3S and fy32 / s Achieving the best result is a subjective decision, an attenuation of 6dB versus a flat one However, frequency response seems adequate.

In addition to the embodiments described and illustrated above Of course, other configurations of the invention are also possible. So can the circuits used e.g. be digital by using e.g. digital memories instead of the delay lines described above used. Instead of receivers that have a tuner, IF amplifier, etc. included, video monitors can also be used. The invention can also be used with many other frequency responses or implement response functions of filters.

3431347

The chrominance channels can contain interpolation filters, which differ from the filter used in the luminance channel. So one can choose p = 0 for the color mustache filter, because the spatial frequency sensitivity of the eye is less sensitive to the chromaticity than to the luminance, as reflected in the smaller bandwidth of the I and Q color television signals relative to the Y signal. In this way, the circuit complexity required can be reduced.

- blank page -

Claims (23)

Claims 1. Arrangement for generating a continuously scanned television picture from line signals which correspond to a picture, which is scanned through a raster in which lines of "even" sub-images have a vertical one Are spaced 2S apart and are interleaved with lines of "odd" fields in such a way that they are opposite them have a distance S, characterized by: a filter device (1252) for spatial frequency filtering of the signals (Y) in a vertical direction with a Filter that has an output signal for each distance level and thereby a large number of filtered line signals at the same time generated for every line of the incoming signal; display means (1274, 1282, 1284, 1240) coupled to the filter means for displaying the plurality to reproduce filtered line signals in adjacent lines of the image raster. - 2 - 2. Arrangement according to claim 1, characterized in that the filter device (1252) comprises: means (1254-1260) for interrogating the line signals with a plurality of taps which are separated from one another according to the spatial frequency 1 / S; a multiplier (1269-1272) associated with each of the taps is coupled to the signals appearing at the taps, each with a predetermined To multiply factor, which can also be equal to 0; a summing device (1273, 1276) connected to the outputs the multiplier is coupled to to deliver the multitude of filtered line signals at the same time, and that the filter means the signal components the room frequency + 1 / 2S attenuates to the value 0. 3. Arrangement according to claim 2, characterized in that the filter device (1700) at least a part of the signal components between the spatial frequencies 0 and + 1 / 2S are amplified to an amplitude that is higher than the signal component of spatial frequency 0, while in the area between the spatial frequencies 0 and + 1 / S the same image phase is retained. 4-, arrangement according to claim 2, characterized in that the multiplier (1316, 1318, 1320, 1322) is coupled to each of the taps by that of each of the taps to multiply the detected signal by a constant each, and that the constants are chosen are that a shortened (sin x) / x distribution results. 5. Arrangement according to claim A , characterized in that the (sin x) / x distribution is broken off behind the second zero of the positive abscissa and behind the second zero of the negative abscissa, so that the con- stanten the main lobe, the first positive minor lobe and the first negative minor lobe of the (sin x) / Fill distribution, 6. Arrangement according to claim 4, characterized in that the number of taps is 5. 7. Arrangement according to claim 4-, characterized in that the number of taps is 7. 8. Arrangement according to claim 4-, characterized in that the number of taps is equal to 5 or equal to 7 and that the filter signals at spatial frequencies between 1 / 3S and 2 / 3S attenuates by an amount greater than 6dB to eliminate the apparent quality of the image to improve the concealment of the adjacent band. 9. Arrangement according to claim 4-, characterized by a such a dimensioning of the constants of the multiplier circuit that they proceed from the central tap decrease monotonically according to the main lobe of a (sin x) / x distribution. . 10. Arrangement according to claim 2, characterized in that that the interrogator is a tapped delay device (1254, 1256, 1258, 1260) has, which is coupled to a source of the video signal, which appears to be interlaced in a field-wise manner, in order to receive the Taps make delayed versions of the video signal appear, which increases in their delay distinguish integer multiples of a line scanning period of the video signal; that the multiplier (1269, 1270, 1271, 1272) is coupled to the taps to increase the amplitude of the delayed video signals in a predetermined manner so that one of the number of .1 taps the same number of measured signal copies is delivered; that said summing means (1273, 1276) summing (1273, 1276) whose number is equal to the sum of 1 plus that number of rows between each p _aa r each incoming partial image to be interpolated from lines, each of said Stunniierschaltungen a receives another group of said sized signal copies to simultaneously generate lines of interpolation filtered video signals, the number of said lines generated at any point in time being equal to the number of summing circuits; parallel to serial converting means (127 * 0 coupled to the summing means for converting the interpolation-filtered video signals appearing at the same time one after the other in a predetermined Pass on time sequence; that the playback device (1282, 1284, 1240) with coupled to the parallel / serial converter is to convert the interpolation-filtered video signals supplied one after the other in one of the mentioned To reproduce time sequence corresponding sequence of different vertical positions. 11. Arrangement according to claim 10, characterized in that the tapped delay device (1254 ·, 1256, 1258, 1260) a cascade connection of several delay lines is. 12. The arrangement according to claim 1, characterized in that the delay lines (1254, 1256, 1258, 1260) are CCD delay lines are. 13 ·. Arrangement according to claim 11, characterized in that the delay lines (1254, 1256, 1258, 1260) are digital memories. 5 - 14 ·. Arrangement according to claim 12, characterized in that . that the multiplier (1269, 1270, 1271, 1272) consists of attenuators. 15. Arrangement according to claim 14, characterized in that that the multiplication factor for a number of the delayed video signals which is smaller than the first Multiplicity, can be equal to 1 and that part of the multiplier which uses this multiplication factor Multiplied by 1, it is simply a senior body. 16. The arrangement according to claim 10, characterized in that the parallel / serial conversion device a Changeover switch (14-70) which operates at high speed between a first and a second output the summing device switches. 17. Arrangement according to claim 10, characterized in that that the summer (1276) has inverting inputs. 18. Arrangement according to claim 10, characterized in that that the parallel / serial converting means comprises time pressing means (1274) to sequentially to generate time-pressed signals with twice the line frequency. 19. Arrangement according to claim 18, characterized in that that the display device (1282, 1284-, 124-0) a has imager (1240) scanned at twice the line frequency. 20. The arrangement according to claim 10, characterized in that the image display device (14A0, 14-38, 14-4-4-, 14-80, 14-92) has a deflection device (14-4-4-), a full deflection of the scanning of the image in - ο - To effect vertical direction at a speed that is less than the line speed, and an auxiliary vertical deflector (11-92) to the scanning at high speed by a relatively distract small measure. 21. Arrangement according to claim 10, characterized in that that the tapped delay device (1710, 1712) has a first and a second 1H delay line (1710 and 1720) and with taps at the input and output of the first delay line and is provided at the output of the second delay line for supplying an undelayed video signal, a video signal delayed by 1H and a video signal delayed by 2H; that the multiplier has a first and a second attenuator (1714 and 1722) for Attenuation of received signals by a factor p which is a predetermined fraction, the first Attenuator for receiving the undelayed video signal connected to a p-damped instantaneous To deliver video signal, and wherein the second attenuator for receiving the delayed by 2H Video signal is connected to provide a p-attenuated 2H delayed video signal; that the multiplier also has a third and a fourth attenuator for attenuating received signals to a fraction 1 / ^, where the third attenuator is connected to receive the video signal delayed by 1H, attenuated by -12dB Deliver 2H delayed video signal; that the multiplier also has a fifth attenuator which attenuates to the fraction i / 2-2p (1716), which is connected to receive the video signal delayed by iH, to a main deliver attenuated 1H delayed video signal; that the summing device has a first summing switch device (1724-) coupled to the third and fourth attenuators for receiving the -i2dB attenuated iH delayed video signal and the ~ 12dB attenuated 2H delayed video signal to produce a first suramen video signal, and in that the summing means further comprises a second summing circuit (1726), the p-damped for receiving the undelayed video signal, the main damped 1H-delayed video signal and the p-damped ■ 2H-delayed video signal connected to. generate a second composite video signal. 22 · Arrangement according to claim 21, characterized in that that the p-attenuated instantaneous and the 2H-delayed video signal inverting inputs of the second Summing circuit (1726) are supplied. 23. The arrangement according to claim 1, characterized in that it is implemented in a color television set that is designed to receive television signals that are representative of the interlaced raster scans for image reproduction in continuous scanning, the filter device comprising: a luminance interrogator (1254, 1256, 1258, 1260) that is used to receive a luminance component (Y) The television signals are connected to this component at a variety of luminance taps which are spaced apart from one another according to a spatial frequency 1 / S, where S the distance between the raster lines of a partial image and the spatially adjacent raster lines of a temporally adjacent partial image; a luminance multiplier device (1269, 1270, 1271, 1272) associated with each of the luminance taps is coupled to the ars diesea taps removed signal with one constant each multiply; - 8th - · ■ "O ■" " a luminance summing device (1276, 1273) with the luminance multiplier is coupled to their output signals in a in such a way that signal constituents of a spatial frequency, viewed in the vertical direction, of -1 / 2S can be damped to 0; a chrominance interrogator adapted to receive a chrominance-representative component of the television signals is designed to remove this component at a large number of taps, which accordingly the spatial frequency 1 / S are separated from each other; a chrominance multiplier coupled to each of the chrominance taps for increasing the amount of the The signals taken from the chromaticity taps have to be multiplied by a constant j a chrominance summing device associated with the Chroma multiplier is coupled to sum their output signals in such a way that signal components are one in the vertical direction spatial frequency seen attenuated from ^ 1 / 2S to 0 will, and that at least one of the chrominance taps is different from the plurality of luminance taps and that the chrominance constants differ from the luminance constants so that the Luminance filtering and chromaticity filtering have different characteristics.